cradek: the pivot length is working great - did you change something?
I am working on the orthogonal compensation now - the last thing
checking for libusb... checking for libusb-config... false
checking for random stuff to make you feel better... failed
checking for libexpat... -lexpat
stustev: heh, nope, I didn't do anything
the only thing I did was update halui so it has pins that show all three tool "lengths"
it seems to be working just like I want
I don't know if I will use the tool lengths but it seemed as if they should have been updated
the buildbot hasnt gotten any email from the commit list since June 19
i'm still getting it
seb_kuzminsky: commits have been going wild...
what address does it use? I can check on it.
stustev: yes if nothing else, you could use that to display them on something.
seb_kuzminsky: "Delivery was disabled by the user via their personal options page."
the only other two options are "by list administrators" and "due to excessive bounces"
but it does say disabled by user for buildbot.
I can reenable it if you want
maybe somebody hax0red my buildbot
i'll do it (if i can remember the passwd...)
huh, 46 people get commit emails - that's more than I expected.
stustev: do you have new git checkouts yet?
yes - on two boxes - worked perfectly and fast
* seb_kuzminsky shrugs
my only complaint is the 'axis jog' is not working yet :)
I have only heard positive things about the git switch so far, except for from one :-)
stustev: "not working yet" is a bit of an understatement...
dude i love it! :-)
would 'not even started' be more accurate?
I'll get used to it - still learning the tricks.
seb_kuzminsky: I see the mails are enabled now
yeah i just poked it
strange that it got disabled
yeah, I expected to see bounces - this was a surprise
change your password for good measure?
now i need to teach it to parse the new commit emails, and teach the slaves to check out from git
stustev, no, it's been started
I think I'm done changing the emails for a while now...
good idea, new passwds all around...
does it have to do anything except pull when it gets one?
oh my - can hardly wait
stustev, you have started the spec, haven't you?
cradek: well it has to know what branch to pull from
cradek: any chance you want to run the buildbot-poker in the git post-commit hook?
seb_kuzminsky: oh true. that's easy - in the subject
the axis jogging spec, of course
can't write code until we know what it's supposed to do
what does it have to do?
the spec is 'jog the axes with a jog wheel or button'
too vague. next!
stustev: including incremental jogs
see, it's changing already!
SWPadnos: actually I think that is the full spec :-)
well, at least it's been started
the spec is 'jog the axes with a jog wheel or button and incremental jogs'
SWPadnos: it should work like jogging currently works on trivkins machines, except now it should be in world mode on kins machines
ok, like jmk's big diagram?
yes that was what jmk's diagram was about
also, he made text notes at fest
I should read them
I didn't see the diagram - drat
at next fest :)
the next fest is scheduled in 30 days :)
the one he/we started at the Indian or Thai restaurant last year
stustev: there's probably still a printout in that room ... (with my power strip and network cable!)
sometime I'll trade your keys for that stuff! hahaha
that is the one advantage of flying (other than the travel time) - less stuff to leave
(seriously, I still have the keys)
what does your power strip and network cable look like - they are probably in my office now
cradek: you and Matt S
it's completely unimportant, don't worry, just making a joke
in fact I found a power strip in the basement just tonight and threw it in my laptop bag - problem solved.
it's the coffee-shop-power-distribution-device
stustev: I can ship them back to you.
what color is your network cable?
I think it was yellow
don't worry about the keys - the time you are here they will be useful
don't you dare ship me a stupid network cable :-)
I have a LOT of yellow cable - ALL of our patch cables are yellow - you can have one - no problem - I won't ship it overnight freight collect :)
darn, I thought I had that pdf diagram, but I don't
let me check the other computer
that pdf is online, lemme find it
I thought the culprit might be lurking :)
[03:15:04] <jmkasunich> http://jmkasunich.com/pics/emc2-motion-dataflow.pdf
[03:15:05] <SWPadnos> http://jmkasunich.com/pics/emc2-motion-dataflow.pdf
ah, I was too slow, I found it too
looking at it - doesn't help me much YET
the text doc is in cvs^H^H^H git
cradek: i think the git repo's post-receive hook should call the git_buildbot.py script distributed with buildbot, and that should do it
it's another thing to set up, but i think it's probably easier than parsing the email
if all you need is the branch, it's the first thing on the subject line
can you give me a url for git_buildbot.py?
[03:19:22] <seb_kuzminsky> http://github.com/djmitche/buildbot/blob/3a95b1107d23a0baf06f0075319ec6b774291690/contrib/git_buildbot.py
or it's in the buildbot package
cradek: other handy things to get would be the committer and the revno of the commit
so if the commits stack up while the buildbot is busy, it can properly attribute changes to committer
* seb_kuzminsky looks at the git_buildbot script
wow, 300 lines to communicate that stuff to the buildmaster?
yeah, it's sure big
argh, python now
every single interpreted language...
I assume I have to change the master = "localhost:9989" line?
yes, or give it --master on the command-line
if you dont want to install python i can update the email parser script i wrote
python is no problem - done already
the buildmaster url is emc2-buildbot.colorado.edu:51332
port 51332, good joke eh?
don't thank me until we see it work :-)
I must have failed leet, dunno what 51332 is supposed to mean. "sleet"?
I would have figured that out any second
I already had it figured out, I was just testing you
yeah, and it was my next guess
anyone mind if i disable the 2.2 builds?
anyone other than paul, i mean
emc 2.2 you mean?
yeah, the 2.2 branch
I'm still using it (not that users = needs to be in buildbot)
if it's still getting commits i should be in the buildbot
it doesnt cost any extra to keep it there, it just clutters the waterfall page a bit
I bet few devs will be doing anything with 2.2, and I don't think we intend to have another 2.2.x bugfix release
that's my thinking too
gps + google maps is amazing
I recently discovered that my "phone" does both of those things
in places where I was walking on a visible path (satellite photos), the track is within about 30 feet of the path
however, the altitude data from the GPS is a lot worse
the first map I posted (which I've since deleted) had lots of up and down that wasn't real
I ran the lat-long track thru an online thing that looks up the actual elevation of the ground, and the new elevation track is much better
* seb_kuzminsky cringes and waits for someone to commit
EMC: 03cradek 07master * re7161ad092a5 10/README: As often as not, people building from source want a sim build.
seb_kuzminsky: ImportError: No module named twisted.spread
nah, it failed
twisted python rocks
what distro are you running on the git host?
how do you say "perl -C" in python?
there's got to be a port for twisted
there are 14 py-twistedSomething
see you guys later
cradek: do you want me to do the script for email parsing?
let me flail for a while longer first
this seems like a better solution
an alternative to that git_buildbot script is to shell out to "buildbot sendchange" directly
it takes command-line arguments specifying the revno, branch, etc, stuff that's probably command-line arguments to git's post-receive hook
of course i bet buildbot needs twisted...
does freebsd ports do dependency tracking between packages?
seb_kuzminsky: I got it to run, and I think I told it the right thing - it should have triggered
yes, but my buildmaster config is broken
oh good :-)
ok try it again
or maybe i can, hold on
I ran it and typed '274bc6f02b747956889f4580dda39f5be55db97e e7161ad092a56b7894ee42a4f75d7c2aa26853b1 refs/heads/master' on stdin
i see you saying that
looks like it took
all the trunk buildslaves are waiting for 1 minute to see if you're going to commit again
that's kind of a cvs-ism, that delay should probably shrink way down now that we're using git
ha, they shouldn't have to do that anymore, should they?
here they go...
now your pythons are broken
welcome to my world
I'm surprised it's trying to use cg - I thought that was discontinued
i thought so too
there's no cogito package in hardy or dapper
i bet it's the older version of buildbot on hardy & dapper vs lenny
seb_kuzminsky: if you need newer git on dapper, install the linuxcnc git1.6 packages from the emc2.3 repo
I need to get to bed - hope you get it soon - if you need anything from me, send me an email.
ok thanks cradek, that post-commit hook helped a lot
cradek: i got git 22.214.171.124 from dapper-backports
EMC: 03cmorley 07master * r3617c8422348 10/src/emc/usr_intf/pncconf/pncconf-help/ (help-load.txt help-welcome.txt): Add help pages
EMC: 03cmorley 07master * r120181e860d8 10/src/emc/usr_intf/pncconf/ (pncconf.glade pncconf.py): Update steptest / add helpwindow
EMC: 03cmorley 07master * rb69ecc965a22 10/ (77 files in 18 dirs): Merge branch 'master' of ssh://email@example.com/git/emc2
micges1 is now known as micges_plasma
EMC: 03jepler 07v2_3_branch * rdb12e7776cb4 10/tcl/bin/halshow.tcl: Fix the watch display - scientific notation
jepler: what do you think about this component: http://www.pastebin.ca/1471045
it's more of a "bus" component than an and array
but it eliminates 20 ands from my hal file
it looks like it should work, the only question I'd have is what to call it
and2_array might be fine
that component should be usable until hal design allow better solutions
out of curiosity, what do you use that for?
Instead of incorporating every component that's written to suit the needs of one person, I'd rather find a way to make it even easier to use custom .comp files
sudo comp --install seems pretty easy
the coolest thing would be if you can 'loadrt mycustomcomp' and it would look in the working directory for mycustomcomp.comp and build it if necessary...
hmmm. I can't tell if you'r ejoking :)
no, I'm pretty serious
I wouldn't mind if I could hide it from users that comps have to actually be built by the C compiler and all that junk
that would be cool, though wouldn't it eliminate half the reasons for having emc_module_helper?
yum, I just got fresh raspberries from the garden
that's certainly one of the open questions..
if not all of them
if not that, maybe a system directory full of .comps that can be rebuilt with one command, and rebuilt automatically when a new emc2 package is installed
I'd be in favor of that level of automation only if all the compilers and tools needed for comp are included in a base install
yes, that's part of my happy place vision
brought on by fresh raspberries?
can someone please try 2.3 with these commits cherry-picked: 2a019987bb35c9761750103afa55b234b184fba2 4cb0cd53921a5fed2710a2030849717f4355cbab
i'd like to include those changes (they lower CPU use in axis dramatically, particularly when showing the big dro)
but I've only tested them on one system
is sim OK?
after checking out v2_3_branch, just git cherry-pick each of those revisions
ok, let's see how long it takes me to figure out how to do that
argh, it conflicts .. hmm
jepler: what values of cpu use you have before and after ?
micges: I don't remember the specific values
I've check that change but it was about 5% change
SWPadnos: don't bother trying to cherry-pick those changes .. I'll prepare a patch that works against 2.3.
ok - I see the conflict
should apply to v2_3_branch with git-am: http://emergent.unpy.net/index.cgi-files/sandbox/axis-display-optimizations.mbox
hmmm but I spot something that's still not quite as intended..
SWPadnos: it'd still be nice to know if that works for you, though
bbl, I have to head for the office
did you more or less have to keep the changes, or was there any real merging involved?
SWPadnos: about and2_Array: I've wrote it to link my front panel of machine iwth many buttons to hal and with option to disable part of panel
so I tried to cherry-pick, which didn't work due to the merge conflict
then I edited the couple of spots that showed conflicts (with the <<<<< markers)
then I tried the patch, which failed in a couple of spots
so I figured I'd start over with it and removed axis.py (using rm)
then tried to git pull, and got a message "You are in the middle of a conflicted merge"
when you fix a merge conflict, you have to say so by git-adding that file
I didn't actually fix it, I just made it pass the syntax check :)
in the conflicts I saw, it looked like you wanted to take the empty version
the other half was related to rotation support
other side, that is
there were only two conflicts, and they were both very short sections, maybe 5 lines, dealing with fonts
that was after the chery-pick
fonts? hm, that's not what I remember seeing
I wonder if I gave the wrong commit ids
I only did the first one
cradek: did you say we had a choice in whether to report merges on -commit / cia? or did I dream that?
because I'm wondering whether the right decision is not to show them..
I don't know of a way to do that offhand
git log has a "no merges" flag
of course if there is a git command that discriminates them I could use it
if the thing that decides what to print is basd on git-rev-list or git-log, you can try --no-merges
yes it is, I'll add that
micges: are the hm2 stepgen bugs you reported earlier all gone now?
it looks like yes
machine is working about week
i was unable to reproduce the hm2 encoder velocity NaN bug you reported, can you give more details on how to reproduce?
all bugs was gone but only one: 1 of 5 boots of pc make direction of one axis inverted, like cables was swapped
but after applying new firmware everything works correct
seb: I will messing with NaN in next week then I'll get some details
seb_kuzminsky: thanks for working on buildbot
oh wait, on thursday I'll be fixing THAT buggy machine on which bugs showed up
seb: but I've learned how to get usefull data so will be ok
jepler: i just do it so you'll keep posting cool zenbot pics
i just need to figure out what to do with dapper now...
seb_kuzminsky: yeah, too bad we're supporting dapper for emc2.3
(hm, I'm a bit surprised that buildbot doesn't abbreviate commit ids for the grid display .. it would be a huge improvement)
seb_kuzminsky: how did you happen to choose buildbot 0.7.9? I see hardy's version is 0.7.6 and the current is 0.7.10p1
ah, the joy of git revnos ;-)
the buildmaster is running interpid, which has 0.7.9
dapper's I see
what is it about dapper that doesn't work with current buildbot?
'python setup.py install --home $HOME' seemed to work
on dapper? cool
i tried to build a .deb, but maybe i should just install by hand
I can't run it, I don't have python-twisted-* installed..
* jepler begs his sysadmin to install it
hmm the testsuite prints at least one error
and more in buildbot.test.test_run
so maybe it's not quite right
seb_kuzminsky: these are the failed tests: http://pastebin.ca/1471320
i think the git fix went into buildbot 0.7.7, maybe that older version does better with dapper
dapper's old twisted
let me get 0.7.7 and try it
seb_kuzminsky: 0.7.7 fails git tests: exceptions.ValueError: invalid literal for int():
.10 has "many git fixes" and "use 'git foo' everywhere"
and 'git foo' (instead of git-foo) is required with the git 1.6 packages I have on this dapper
I have no idea if those other tests are important to the way we use buildbot; how can I test this system as an emc2-building slave? I should be able to do a deb build of v2_3_branch, or a sim build
here's my instructions for setting up a buildslave: http://emc2-buildbot.colorado.edu/buildslave-admin-guide.html
heh, guess i should update that for git
yes, it's a testsuite bug
it doesn't like a version string like 126.96.36.199.469.gdffc
i can't say i blame it
OK, I think that buildbot 0.7.7 + small changes + git 188.8.131.52 should work on dapper
dapper-backports has git 184.108.40.206, but you already packaged 1.6 for dapper right?
seb_kuzminsky: right, cradek did
the packages are in the dapper repo
EMC: 03jepler 07master * r35f6955c4787 10/ (27 files in 3 dirs): remove the abandoned mesa_5i2x driver project per TODO
jepler: did cmorley revert BJT's commit that changes docs to 1.5.3?
no little test commit, that
I suppose I should have tested whether emc still built after that before pushing :-P
jepler: from cmorley's commit: 77 files changed, 31175 insertions(+), 20094 deletions(-)
that was a merge...
the commitdiff isn't helpfull though: http://git.linuxcnc.org/?p=emc2.git;a=commitdiff;h=b69ecc9
cradek: can you elaborate?
I did make a change so we won't get the merge notices
!git diff 35f6955c4787bdaca890e4ba78a1d46f73ed9404 274bc6f02b747956889f4580dda39f5be55db97e -- docs
I ran this command, and it doesn't show differences
ah, it's described here: http://lists.sip-router.org/pipermail/sr-dev/2009-April/001386.html
(those are the revisions of origin/master and "Convert Documents")
it doesn't show differences, so I know that chris morley's commit didn't unintentionally undo those changes
jepler: see the link I pasted, it describes the exact same issue some other guys had
he pulled, then did his changes
in the mean time master was incremented by BJTs commit
without doing this detective work I'm not sure how to tell whether the result of the merge is what was right -- for instance, if you merge with --strategy=ours, you throw away all the changes in the other side of the merge..
when he tried to push, he had a stale checkout, so he had to pull
if he would do "git pull --rebase" that would undo his changes, get the changes from master, then apply his changes locally
then a "git push" would be with the revisions in the expected order
EMC: 03jepler 07master * r4a3497da7251 10/src/ (Makefile Makefile.inc.in): finish removing hal_5i2x
alex_joni: on main page linuxcnc.org there is no login section , on others pages it is
micges: I know, I disabled it
we get lots of spammers registering on the first page
seb_kuzminsky: oh, I should make it clear that what I ended up using was 0.7.10p1 unmodified
jepler: on my machine axis cpu usage didn't change between preview and dro, but I have xorg cpu=32% on preview and 22% on DRO (if that helps you)
the difference is dramatic on my system. Before, Xorg 36% axis 4% while running flowsnake in a maximized window on a 1600x1200 screen. After, Xorg <2%. Of course, this is a "sim" machine with nvidia opengl installed..
on trunk xorg=30%, on 2.3.2 xorg=60% so change is big
ok, the relative numbers help me understand what you meant
* alex_joni tries to rebase joints-axes2 against master
wonder if it was right, even back then
* cradek vaguely wonders how so many people are managing to have broken opengl lately
I wonder if I really borked the rebase :)
if I checked out the joints-axes2 branch
and I did a rebase against master
I now have all the commits that happened on master after joints-axes2 has been branched
how can I check what will be pushed?
run gitk and you'll see them
that's what I'm doing, but I can't say I'm "getting" it
I did git branch ja origin/joints_axes
git branch ja origin/joints_axes2
git checkout ja
the unstaged stuff is from git changing CRLF to LF (if I understood it correctly)
git rebase master
but then I got a conflict
jepler: I resolved the conflicts
in fact, a lot of them
ok, after you resolved them did you When you have resolved this problem run "git rebase --continue".
initially I had some conflicts, then it errored out, that it can't patch
er, did you "git rebase --continue"?
and that it will use the regular merge
(it looks like you didn't, because gitk shows that you have changes that haven't been checked in)
those are things git changed
I got a "Warning: CRLF line endings will be replaced with LF"
anyways, I'll try again tomorrow.. gotta run now :/
early trip to the airport tomorrow morning
good night all
ugh, have fun
if you have changes that are not committed, then you haven't "resolved the problem" and "run git rebase --continue"
(you have to "git add" the file after you resolve the conflict manually)
EMC: 03jepler 07v2_3_branch * r5e22d76bba36 10/src/emc/usr_intf/axis/scripts/axis.py: in big dro, avoid doing some expensive operations every redraw
EMC: 03jepler 07v2_3_branch * rafd4e5ec2856 10/src/emc/usr_intf/axis/scripts/axis.py: don't redraw the big dro when showing plot window, and vice versa
EMC: 03jepler 07v2_3_branch * re3f44a2fcbdf 10/src/emc/usr_intf/axis/scripts/axis.py: this should short-circuit the redraw
EMC: 03jepler 07master * rd069a1e4ee93 10/src/emc/usr_intf/axis/scripts/axis.py: this should short-circuit the redraw
Roguish_ is now known as Roguish
KimK is now known as KimK_afk